Crystallization is one of the most used separation processes in the chemical industry to produce high-value-added products, and its success is dependent on controlling properties such as kinetics, crystal size distribution (CSD), shape, and polymorphism. To optimize the process, the Food and Drug Administration (FDA) encourages the development of small-scale online and in-line techniques to obtain real-time data. This work proposes using the FBRM equipment for monitoring the crystallization process by efficiently converting chord length distribution (CLD) measured into CSD, which are used to estimate kinetic parameters. It was possible to train artificial neural networks (ANN) to covert CLD in CSD using different training methods (Levenberg-Marquardt, Bayesian Regularization, and Scaled Conjugate Gradient) and test them to obtain optimized networks that had a maximum deviation of 15% compared to the DTC obtained in Malvern. To ensure the process reliability, four crystallization assays (A-D) were monitored by the best ANN, which allowed the prediction of experimental kinetic parameters through the Method of Moments and inferences about the process, compared to documented experiments. The nucleation kinetic parameters (n from 1.15 to 2.05 and kn from 1,19E + 11 to 3,52E + 12) and growth kinetic parameters (g from 1.6 to 2.00 and kg from 4,96E-06 to 1,94E-04) were obtained, approaching experiments described by other authors under similar conditions. The use of ANNs to analyze FBRM data has shown efficiency in attributing physical meaning to CLD data converted into CSD. Through this method, it is possible to estimate kinetic parameters at each instant of the crystallization process, paving the way for adjusting process parameters in real-time to achieve the final product's required quality by regulatory agencies.